Infrared and nebular lines provide some of our best probes of the physics regulating the properties of the interstellar medium (ISM) at high-redshift. However, interpreting the physical conditions of high-redshift galaxies directly from emission lines remains complicated due to inhomogeneities in temperature, density, metallicity, ionisation parameter, and spectral hardness. We present a new suite of cosmological, radiation-hydrodynamics simulations, each centred on a massive Lyman-break galaxy that resolves such properties in an inhomogeneous ISM. Many of the simulated systems exhibit transient but well defined gaseous disks that appear as velocity gradients in [CII]~158.6$\mu$m emission. Spatial and spectral offsets between [CII]~158.6$\mu$m and [OIII]~88.33$\mu$m are common, but not ubiquitous, as each line probes a different phase of the ISM. These systems fall on the local [CII]-SFR relation, consistent with newer observations that question previously observed [CII]~158.6$\mu$m deficits. Our galaxies are consistent with the nebular line properties of observed $z\sim2-3$ galaxies and reproduce offsets on the BPT and mass-excitation diagrams compared to local galaxies due to higher star formation rate (SFR), excitation, and specific-SFR, as well as harder spectra from young, metal-poor binaries. We predict that local calibrations between H$\alpha$ and [OII]~3727$\AA$ luminosity and galaxy SFR apply up to $z>10$, as do the local relations between certain strong line diagnostics (R23 and [OIII]~5007$\AA$/H$\beta$) and galaxy metallicity. Our new simulations are well suited to interpret the observations of line emission from current (ALMA and HST) and upcoming facilities (JWST and ngVLA).

Accurate models of radiative cooling are a fundamental ingredient of modern cosmological simulations. Without cooling, accreted baryons will not efficiently dissipate their energy and collapse to the centres of haloes to form stars. It is well established that local variations in the amplitude and shape of the spectral energy distribution of the radiation field can drastically alter the cooling rate. Here we introduce deepCool, deepHeat, and deepMetal: methods for accurately modelling the total cooling rates, total heating rates, and metal-line only cooling rates of irradiated gas using artificial neural networks. We train our algorithm on a high-resolution cosmological radiation hydrodynamics simulation and demonstrate that we can predict the cooling rate, as measured with the photoionisation code CLOUDY, under the influence of a local radiation field, to an accuracy of ~5%. Our method is computationally and memory efficient, making it suitable for deployment in state-of-the-art radiation hydrodynamics simulations. We show that the circumgalactic medium and diffuse gas surrounding the central regions of a galaxy are most affected by the interplay of radiation and gas, and that standard cooling functions that ignore the local radiation field can incorrectly predict the cooling rate by more than an order of magnitude, indicating that the baryon cycle in galaxies is affected by the influence of a local radiation field on the cooling rate.